Robert Axtell, Jeffrey Dean, Joshua Epstein, George Gumerman, Alan Swedlund

Paper #: 98-10-094

An agent-based computational model of Long House Valley, in northern Arizona near Monument Valley, is described and demontrated. The model, that runs from about A.D. 400 to 1400, consists of artificial adaptive agents (households) who inhabit a digitized version of the Long House Valley landscape. A detailed paleoenvironmental record exists for Long House Valley, based on alluvial geomorphology, palynology, and dendroclimatology. This permits accurate quantitative reconstruction of annual fluctuations in the Valley's potential agricultural production. Agents are given rules of behavior for determining agricultural and residential locations, as well as for reproduction and mortality. Each run of the model generates a unique history of population, agricultural output, and settlement patterns. Results from a large number of runs are then compared to the extensive archaeological data on this area, data resulting from a 100% survey supplemented by limited excavations.